Coinbase is deploying machine learning, AI and generative AI to automate customer interactions and combat fraud.

The crypto exchange and trading platform is working with Amazon Web Services to deploy AI and gen AI technology for internal and external chatbots, Alex Reeve, director of product management, said May 15 during the AWS Financial Services Symposium in New York.

“Machine learning and AI has been injected in every part of our business,” Reeve said.  

Chatbot 

Coinbase deployed a gen AI-driven chatbot within its app to solve customer queries as the first line of interaction in January 2023, Reeve said. 

Since launching the chatbot, the company said it has recorded the following efficiencies: 

• Nearly 64% of all customer inquiries are solved by the chatbot, up from 17% in 2023; and 

• 5 million employee hours have been saved. 

“We operate in over 100 countries and multiple language regions,” Reeve said. “The chatbot has aided us in serving our customers without hiring a big customer relationship team.” 

The bot answers queries about adding money, pending transactions and login issues, Reeve said, noting that if a query escalates it is transferred to a human agent. 

Customer experience 

Coinbase also uses AI to make customer experience more personalized, Reeve said, adding that the app provides customers with market updates according to their interests and holdings. 

The company has also deployed an internal AI-driven virtual assistant for its customer relationships team, Reeve said, adding that the AI assistant helps customer relationship employees solve queries quickly and more efficiently. 

The company did not disclose the specific AI models, like OpenAI, Gemini or Anthropic, are being used for its chatbots. 

Anti-fraud efforts 

Coinbase also has launched machine learning to combat fraudulent transactions and account takeovers, Reeve said, adding that machine learning has been a part of its operations for nearly a decade. 

The machine learning model can detect fraud in real time and urges users to proceed with caution before making transfers to unknown entities, Reeve said. 

Despite deploying machine learning for fraud detection, Coinbase recently fell prey to cyber-attacks. 

Hack attack 

While Coinbase has been growing and was added to the Standard & Poor’s 500 index May 13, it faced a cyber hack on May 11, resulting in the loss of sensitive information of less than 1% of its clients, according to a May 15 release from the company. 

Cyber criminals bribed and recruited a group of rogue overseas support agents to steal customer data to facilitate social engineering attacks, the release stated. 

The cyber criminals asked for a $20 million ransom from the company, but instead of agreeing to the demands, Coinbase decided to put a $20 million bounty on the criminals, according to the release. 

“No passwords, private keys, or funds were exposed, and Coinbase Prime accounts are untouched,” the company wrote. “We will reimburse customers who were tricked into sending funds to the attacker.” 

Cyber criminals bribed and recruited rogue overseas support agents to pull personal data on <1% of Coinbase MTUs. No passwords, private keys, or funds were exposed. Prime accounts are untouched. We will reimburse impacted customers. More here: https://t.co/SidVn59JCV

— Coinbase 🛡️ (@coinbase) May 15, 2025

The company is under investigation by the Securities and Exchange Commission for misrepresenting its total customer base in its initial public offering prospectus filed in 2021. Chief Legal Officer Paul Grewal told Bloomberg on May 15 that the lawsuit “is a holdover investigation from the prior administration” and Coinbase is cooperating with SEC to close the lawsuit.